Establishing workflow trust using provenance information

Workflow forms a key part of many existing Service Oriented applications, involving the integration of services that may be made available at distributed sites. It is possible to distinguish between an \abstract" workflow description outlining which services must be involved in a workflow execution and a \physical" workflow description outlining the instances of services that were used in a particular enactment. Provenance information provides a useful way to capture the physical workflow description automatically - especially if this information is captured in a standard format. Subsequent analysis on this provenance information may be used to evaluate whether the abstract workflow description has been adhered to, and to enable a user executing a work°ow-based application to establish \trust" in the outcome. An analysis tool that makes use of provenance information to assist in evaluating trust in the outcome of a workflow execution is presented. The analysis tool makes use of a rule-based engine, supporting a range of queries on the recorded information by one or more workflow enactors. The results of the analysis tool on a particular workflow scenario are presented, along with an experiment demonstrating how the analysis tool would scale as the granularity of the recorded provenance information was increased